Release cortex 0.7.0 by juliusclausnitzer · Pull Request #1385 · cobaltcore-dev/cortex

juliusclausnitzer · 2026-10-19T09:19:17Z

Release cortex 0.7.0

⚠️ Breaking Changes

History CRD removed: The cortex.cloud_histories CRD and all associated controllers/tasks have been deleted. Any tooling relying on kubectl get histories or the history client library must migrate to spec.cortex.explanation on the VirtualMachine CRD.
CRD schema changes: Decision and VirtualMachine CRDs updated with new explanation and intent fields.

Features

Improve virtual machine creation logging — Replaced the opaque boolean createVM gate with a switch that logs a specific skip reason (including instanceUUID) for each unmet precondition, plus logs successful VM create/update events for end-to-end traceability.
Populate cortex intent and explanation on VM spec (#1373) — Bumped vm-operator to v0.2.8 and writes the new Explanation and Intent fields into the VirtualMachine CRD spec during scheduling, replacing the now-removed History CRD as the explanation surface.
Prune aged committed-resource reservations (#1382) — The syncer now detects and deletes stale CR-backed Reservations that exceed a configurable max age (capped per run), so the reconciler can rebuild fresh slots rather than accumulating stale ones.

Bug Fixes

Fix broken feature gate configuration in nova bundle — Feature gate keys (committedResourceTracking, virtualMachineCreation, virtualMachineCreationMandatory) were incorrectly nested under a featureGates: block; moved to the correct nesting level.
CRs confirm superseded/expired CRs correctly within CR transactions (#1372) — Superseded/expired CR transitions are no longer added to the poll-watch list, eliminating CortexNovaCommittedResourceTimeoutsTooHigh alerts. Also adds a generation-advance hang guard and wraps reservation deletes in IgnoreNotFound.
Fix scheduling intent set to Unknown always (#1374) — Fixed a bug that caused VM resources to never be generated because intent was always Unknown.
Default scheduling intent to Create for Nova API (#1376) — When intent cannot be determined, it now defaults to CreateIntent instead of Unknown; removes redundant intent derivation that overrode the fallback.
Fix VM status condition being wiped on create path — Snapshots conditions before Create and restores them before the status sub-resource update so the Scheduled condition is preserved.
Use decision intent in VM handler instead of re-deriving from request (#1379) — Reads intent from decision.Spec.Intent (which already incorporates the API's fallback logic) rather than re-calling GetIntent() on the request, fixing initial VM create failures.
Change execution order (#1384) — Corrected execution ordering in the reservations path.

Chores & Refactoring

Remove History CRD (#1375) — Deleted the History custom resource definition, the history client, and all history cleanup tasks across Nova, Cinder, and Manila schedulers. Explanation data is now embedded directly on the VirtualMachine CRD.
Add edge cases to explanation (#1383) — Hardened the plain-text explanation generator with additional edge-case handling.

Documentation

Correct Decision CRD section (#1381) — Clarified that Decision is a transient in-memory CRD (not persisted to etcd); replaced the misleading kubectl get decisions example with kubectl get virtualmachines showing spec.cortex.* fields.

Dependency Updates

Dependency	Old	New
`vm-operator`	v0.1.2-pre	v0.2.9
`kube-prometheus-stack`	92.7.0	92.10.1
`controller-tools`	v0.22.0	v0.23.0
`golangci-lint`	v2.14.0	v2.15.0

Package Versions

Package	Version
cortex	0.7.0
cortex-postgres	0.6.28
cortex-shim	0.1.33
bundles	0.0.107
cortex-placement-shim	0.1.33

Operational Notes

Alert resolution: This release fixes the CortexNovaCommittedResourceTimeoutsTooHigh alert by no longer polling superseded/expired CR transitions.
Stale reservation pruning: CR-backed Reservations are now automatically pruned past a configurable max age — verify the default threshold is appropriate for your deployment.
Migration required: Any tooling using kubectl get histories or the history client library must migrate to spec.cortex.explanation on the VirtualMachine CRD.

Contributors: @PhilippMatthes, @mblos, @SoWieMarkus, @cortex-ai-agents[bot], @renovate[bot], @github-actions[bot]

PR: #1385 | Commits: 20 | Files changed: 91 | Test coverage: 71.4%

Dependencies

Prep PR: Release cortex 0.7.0 #1388 (must be merged before this PR)

Log the specific reason (with instance UUID) whenever VM creation is skipped, and log successful create/update events in the vm handler so resource processing can be traced. Signed-off-by: Philipp Matthes <p.matthes@sap.com>

Signed-off-by: Philipp Matthes <p.matthes@sap.com>

…ons (#1372) Fixes `CortexNovaCommittedResourceTimeoutsTooHigh` alerts when a batch carries a superseded/expired commitment. Also adds a hang guard for cores commitments --------- Signed-off-by: Marcel <156897072+mblos@users.noreply.github.com>

Bump vm-operator to v0.2.8 and set the new CortexSpec Explanation and Intent fields when writing the scheduling result. The intent lookup is moved earlier so it can feed both the spec fields and the status condition. --------- Signed-off-by: Philipp Matthes <p.matthes@sap.com>

Fixes a bug that lead to the virtual machine resources never being generated. Signed-off-by: Philipp Matthes <p.matthes@sap.com>

Default the intent to api.CreateIntent instead of Unknown when it cannot be determined, and remove the redundant intent derivation in the pipeline controller that would otherwise override it with Unknown. --------- Signed-off-by: Philipp Matthes <p.matthes@sap.com>

h.Create writes the apiserver response (with an empty status subresource) back into the vm object, overwriting the Scheduled condition set earlier. The subsequent Status().Update then persisted an empty status. Snapshot the conditions before Create and restore them before the status update. Signed-off-by: Philipp Matthes <p.matthes@sap.com>

Explanation is now added to the vm crd. - adjusted the docs - removed history crd - removed history client - added plain explanation generation Signed-off-by: Markus Wieland <markus.wieland@sap.com>

This PR contains the following updates: | Package | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Adoption](https://docs.renovatebot.com/merge-confidence/) | [Passing](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | Type | Update | |---|---|---|---|---|---|---|---| | [github.com/cobaltcore-dev/vm-operator](https://redirect.github.com/cobaltcore-dev/vm-operator) | `v0.2.8` → `v0.2.9` | ![age](https://developer.mend.io/api/mc/badges/age/go/github.com%2fcobaltcore-dev%2fvm-operator/v0.2.9?slim=true) | ![adoption](https://developer.mend.io/api/mc/badges/adoption/go/github.com%2fcobaltcore-dev%2fvm-operator/v0.2.9?slim=true) | ![passing](https://developer.mend.io/api/mc/badges/compatibility/go/github.com%2fcobaltcore-dev%2fvm-operator/v0.2.8/v0.2.9?slim=true) | ![confidence](https://developer.mend.io/api/mc/badges/confidence/go/github.com%2fcobaltcore-dev%2fvm-operator/v0.2.8/v0.2.9?slim=true) | require | patch | | [github.com/golangci/golangci-lint/v2](https://redirect.github.com/golangci/golangci-lint) | `v2.14.0` → `v2.15.0` | ![age](https://developer.mend.io/api/mc/badges/age/go/github.com%2fgolangci%2fgolangci-lint%2fv2/v2.15.0?slim=true) | ![adoption](https://developer.mend.io/api/mc/badges/adoption/go/github.com%2fgolangci%2fgolangci-lint%2fv2/v2.15.0?slim=true) | ![passing](https://developer.mend.io/api/mc/badges/compatibility/go/github.com%2fgolangci%2fgolangci-lint%2fv2/v2.14.0/v2.15.0?slim=true) | ![confidence](https://developer.mend.io/api/mc/badges/confidence/go/github.com%2fgolangci%2fgolangci-lint%2fv2/v2.14.0/v2.15.0?slim=true) | | minor | | [kube-prometheus-stack](https://redirect.github.com/prometheus-operator/kube-prometheus) ([source](https://redirect.github.com/prometheus-community/helm-charts)) | `92.9.0` → `92.10.0` | ![age](https://developer.mend.io/api/mc/badges/age/docker/ghcr.io%2fprometheus-community%2fcharts%2fkube-prometheus-stack/92.10.0?slim=true) | ![adoption](https://developer.mend.io/api/mc/badges/adoption/docker/ghcr.io%2fprometheus-community%2fcharts%2fkube-prometheus-stack/92.10.0?slim=true) | ![passing](https://developer.mend.io/api/mc/badges/compatibility/docker/ghcr.io%2fprometheus-community%2fcharts%2fkube-prometheus-stack/92.9.0/92.10.0?slim=true) | ![confidence](https://developer.mend.io/api/mc/badges/confidence/docker/ghcr.io%2fprometheus-community%2fcharts%2fkube-prometheus-stack/92.9.0/92.10.0?slim=true) | | minor | --- ### Release Notes <details> <summary>cobaltcore-dev/vm-operator (github.com/cobaltcore-dev/vm-operator)</summary> ### [`v0.2.9`](https://redirect.github.com/cobaltcore-dev/vm-operator/releases/tag/v0.2.9) [Compare Source](https://redirect.github.com/cobaltcore-dev/vm-operator/compare/v0.2.8...v0.2.9) **Full Changelog**: <cobaltcore-dev/vm-operator@v0.2.8...v0.2.9> </details> <details> <summary>golangci/golangci-lint (github.com/golangci/golangci-lint/v2)</summary> ### [`v2.15.0`](https://redirect.github.com/golangci/golangci-lint/blob/HEAD/CHANGELOG.md#v2150) [Compare Source](https://redirect.github.com/golangci/golangci-lint/compare/v2.14.0...v2.15.0) *Released on 2026-10-16* 1. Linters new features or changes - `errcheck`: from 1.20.0 to 1.30.0 - `exhaustive`: from 0.13.0 to 0.15.0 (⚠️ changes) - `gocognit`: from 1.2.1 to 1.3.0 - `golines`: from 0.15.0 to 0.16.0 - `govet-modernize`: from 0.50.0 to 0.52.0 - `loggercheck`: from 0.12.0 to 0.13.0 - `predeclared`: from 0.2.2 to 0.3.0 - `wrapcheck`: from 2.12.0 to 2.13.0 2. Linters bug fixes - `bidichk`: from 0.3.3 to 0.3.4 - `canonicalheader`: from [`a25c71c`](https://redirect.github.com/golangci/golangci-lint/commit/a25c71c521f6) to [`e18e5c5`](https://redirect.github.com/golangci/golangci-lint/commit/e18e5c5a078f) - `gofumpt`: improve deprecation logs - `gosec`: improve concurrency default value - `iface`: from 1.5.1 to 1.5.3 - `paralleltest`: from 1.0.15 to 1.0.16 - `revive`: from 1.17.0 to 1.17.1 - `revive`: the rules `context-keys-type`, `errorf`, `modifies-value-receiver`, `range-val-address`, `string-of-int`, `time-equal`, `time-naming`, `unexported-return`, `unhandled-error`, `var-declaration` are working (less randomly). - `rowserrcheck`: from [`c5f79b8`](https://redirect.github.com/golangci/golangci-lint/commit/c5f79b8a11ba) to [`e0740d5`](https://redirect.github.com/golangci/golangci-lint/commit/e0740d52dee8) - `staticcheck`: from 0.8.1 to 0.8.2 3. Bug fixes - fix: compute custom version hash with module versions - fix: env vars loading (related to `revive`) ⚠️ Changes of the `exhaustive` linter: - The replaced options still exists, and partially works, but they are deprecated. - The options `explicit-exhaustive-switch` and `explicit-exhaustive-map` have been replaced by `explicit-exhaustive`. - The `check` the value `map` has been replaced by `mapliteral`. - The option `ignore-enum-members` (string) has been replaced by `ignore-constants` (slice). - The option `ignore-enum-types` (string) has been replaced by `ignore-types` (slice). - New report messages. - The linter is a now a "slow" linter. </details> <details> <summary>prometheus-community/helm-charts (kube-prometheus-stack)</summary> ### [`v92.10.0`](https://redirect.github.com/prometheus-community/helm-charts/releases/tag/kube-prometheus-stack-92.10.0) kube-prometheus-stack collects Kubernetes manifests, Grafana dashboards, and Prometheus rules combined with documentation and scripts to provide easy to operate end-to-end Kubernetes cluster monitoring with Prometheus using the Prometheus Operator. #### What's Changed - \[kube-prometheus-stack] Update <https://github.com/kubernetes-monitoring/kubernetes-mixin> digest to [`f57ebcd`](https://redirect.github.com/prometheus-community/helm-charts/commit/f57ebcd) by [@renovate](https://redirect.github.com/renovate)\[bot] in [#7382](https://redirect.github.com/prometheus-community/helm-charts/pull/7382) **Full Changelog**: <prometheus-community/helm-charts@kube-prometheus-stack-92.9.0...kube-prometheus-stack-92.10.0> </details> --- ### Configuration 📅 **Schedule**: (in timezone Europe/Berlin) - Branch creation - "after 6pm every weekday,every weekend,before 8am every weekday" - Automerge - At any time (no schedule defined) 🚦 **Automerge**: Enabled. ♻ **Rebasing**: Whenever PR is behind base branch, or you tick the rebase/retry checkbox. 👻 **Immortal**: This PR will be recreated if closed unmerged. Get [config help](https://redirect.github.com/renovatebot/renovate/discussions) if that's undesired. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR was generated by [Mend Renovate](https://mend.io/renovate/). View the [repository job log](https://developer.mend.io/github/cobaltcore-dev/cortex).  Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>

Reservations drift from their VM mapping and settle on invalid hosts over time. The syncer now prunes the oldest committed-resource Reservations past a max age (capped per run) before upserting, letting the reconciler rebuild fresh slots. Signed-off-by: Marcel <156897072+mblos@users.noreply.github.com>

…sted (#1381) ## Summary - **Corrects the "Inspecting decisions and descheduling" section** in `docs/02-external-scheduler-api/02-nova-smart-scheduling.md` to reflect that the `Decision` CRD is a transient in-memory object, not persisted to etcd (as documented in `api/v1alpha1/decision_types.go` lines 126-132) - **Replaces the misleading `kubectl get decisions` example** (which would return no results) with a `kubectl get virtualmachines` command showing the `spec.cortex.explanation`, `spec.cortex.intent`, and `spec.cortex.hostCandidates` fields — the actual persisted scheduling results - **Keeps the `kubectl get deschedulings` example** unchanged, since `Descheduling` resources are persisted and the example is accurate ## Test plan - [ ] Verify the renamed heading anchor (`#inspecting-scheduling-results`) does not break any cross-links in the docs (confirmed: no files linked to the old anchor) - [ ] Confirm the `spec.cortex.explanation`, `spec.cortex.hostCandidates`, and `spec.cortex.intent` field paths match the code in `internal/scheduling/nova/vm_handler.go` - [ ] Confirm the `Descheduling` print columns in the kept example still match `api/v1alpha1/descheduling_types.go` - [ ] Read the section in context to verify it flows naturally between "Closing the loop: descheduling" and the closing paragraph / `## Next` 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: cortex-ai-agents[bot] <279748396+cortex-ai-agents[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

…quest (#1379) ## Summary - Fixes a bug where `VirtualMachineHandler.Upsert()` would fail on every initial VM create because it re-called `request.GetIntent()` which returns an error when Nova doesn't set the `_nova_check_type` scheduler hint (the case for initial placements) - The API layer already handles this gracefully by defaulting to `CreateIntent` and storing it on `decision.Spec.Intent` — the VM handler now reads from that field instead of re-deriving - Updates the test helper `vmTestDecision()` to include `Spec.Intent`, matching the real data flow ## Context This was introduced in #1362 (Create VM resources as part of nova pipeline) and partially addressed in #1374 and #1376 (intent fixes), but the VM handler still had its own `GetIntent()` call that didn't benefit from the API's fallback logic. ## Test plan - [ ] Existing `vm_handler_test.go` tests pass with the updated decision helper - [ ] Verify that initial VM creates no longer fail when `VirtualMachineCreation` feature gate is enabled 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: cortex-ai-agents[bot] <279748396+cortex-ai-agents[bot]@users.noreply.github.com> Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>

Signed-off-by: Markus Wieland <markus.wieland@sap.com>

This PR contains the following updates: | Package | Update | Change | |---|---|---| | [kube-prometheus-stack](https://redirect.github.com/prometheus-operator/kube-prometheus) ([source](https://redirect.github.com/prometheus-community/helm-charts)) | patch | `92.10.0` → `92.10.1` | --- ### Release Notes <details> <summary>prometheus-community/helm-charts (kube-prometheus-stack)</summary> ### [`v92.10.1`](https://redirect.github.com/prometheus-community/helm-charts/releases/tag/kube-prometheus-stack-92.10.1) [Compare Source](https://redirect.github.com/prometheus-community/helm-charts/compare/kube-prometheus-stack-92.10.0...kube-prometheus-stack-92.10.1) kube-prometheus-stack collects Kubernetes manifests, Grafana dashboards, and Prometheus rules combined with documentation and scripts to provide easy to operate end-to-end Kubernetes cluster monitoring with Prometheus using the Prometheus Operator. #### What's Changed - \[kube-prometheus-stack] Update grafana Docker tag to v14.0.1 by [@renovate](https://redirect.github.com/renovate)\[bot] in [#7383](https://redirect.github.com/prometheus-community/helm-charts/pull/7383) **Full Changelog**: <prometheus-community/helm-charts@kube-prometheus-stack-92.10.0...kube-prometheus-stack-92.10.1> </details> --- ### Configuration 📅 **Schedule**: (in timezone Europe/Berlin) - Branch creation - "after 6pm every weekday,every weekend,before 8am every weekday" - Automerge - At any time (no schedule defined) 🚦 **Automerge**: Enabled. ♻ **Rebasing**: Whenever PR is behind base branch, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] If you want to rebase/retry this PR, check this box --- This PR was generated by [Mend Renovate](https://mend.io/renovate/). View the [repository job log](https://developer.mend.io/github/cobaltcore-dev/cortex).  Co-authored-by: renovate[bot] <29139614+renovate[bot]@users.noreply.github.com>

Signed-off-by: Marcel <156897072+mblos@users.noreply.github.com>

github-actions · 2026-10-19T09:19:39Z